Figure 2

Figure 2. Phylogenetic analysis of SARS-CoV-2 sequences based on 1,966 sequences from the NTU-HCH surveillance program in Taiwan and 881 sequences from GISAID (https://www.gisaid.org). Lineages of NTU-HCH strains are annotated...

To gain insights into community transmission and to monitor viral evolution, we deployed a genomic surveillance protocol at National Taiwan University Hospital Hsinchu Branch (NTU-HCH) whereby we performed whole-genome sequencing on nasal swab samples detected by PCR to be positive for SARS-CoV-2 (Appendix 1). To ensure data quality, we submitted genomic data to GISAID (https://www.gisaid.org) only on those sequences that had >98% coverage of the 29,903-bp SARS-CoV-2 target genome. We used the same set of high-quality sequences for tracking the signature mutations in the viral samples (Table 1) and for phylogenetic analysis (Figure 2; Appendix 1, Figure 1). We found 2,405 samples among 5 batches that met the above criterion and this generated 2,043 sequences (84.9% pass rate). We selected 1,966 sequences for GISAID submission (Appendix 1 Table 1).

We analyzed the assembled viral genome sequences (Appendix 1 Table 2) and tracked the lineages and nonsynonymous amino acid changes in the Omicron samples collected during 2022 (Appendix 1 Figures 2, 3). Comparing the later 3 datasets (batches 3–5), we discovered that 3 amino acid changes (open reading frame [ORF]1a: L631F; spike (S): K97E; nucleocapsid; M322I) occurred only after the fourth sequencing batch. The percentage of sequences containing the signatures progressed steadily from 62% in batch 4 to 85% in batch 5. All batch-3 isolates belonged to the BA.1 or BA.2 classification, suggesting that the rapid increase of cases in Taiwan in April and May 2022—from 0 cases/day to ≈100,000 cases/day—came from a strain (BA.2.3.7) that might have been involved in a founder effect.

To construct the framework of the phylogenetic tree, we took 1,966 genome sequences from our study and analyzed them in the global context of 881 GISAID reference sequences (Figure 2; Appendix 2). We then zoomed in and compared the 1,577 Omicron sequences of our study against the 228 Omicron BA.2.3.7 strains from GISAID. Those sequences were reported from 21 countries, including 51 from Taiwan (Table 2). We conducted phylogenetic analysis using the Pango-dynamic nomenclature system (11).

We found evidence that this novel lineage BA.2.3.7 with 3 amino acid changes (ORF1a: L631F; S: K97E; and nucleocapsid: M322I) was circulating dominantly in Taiwan over the study period. Of note, the first BA.2.3.7 strain identified in the epidemic in Taiwan was collected on March 27, 2022, and since that time we detected several genomic changes affecting this Omicron lineage. For example, we noted a new mutation, G1251V (Appendix 1 Figure 3, green line) in the S protein, from April onward, and that particular circulating lineage then rapidly spread across Taiwan.

We acknowledge that our study is limited in that we conducted the genomic surveillance in only 1 medical center; therefore, the observed dominance of BA.2.3.7 might be due to clustering of cases. Of note, while this paper was in preparation, we became aware that several viral sequences with the same signature mutations had been reported in Taiwan. Although the number of cases was relatively small (51) compared with the number of cases we studied, the 4 locations in Taiwan reporting those cases were different from our collection point at the hospital. Thus, this new lineage appeared to be broadly detectable across Taiwan.

Other Asia-Pacific countries have also recently reported a substantial cumulative prevalence of the BA.2.3.7 variant (Table 2). Among the 44 Omicron BA.2.3.7 strains reported from Japan, 2 of the affected persons had travel history to Vietnam and 41 to Taiwan, suggesting considerable silent outward transmission from Taiwan. In contrast, BA.2.3.7 accounts for <0.5% of the sequences reported in either California, USA, or globally. The emergence of Omicron BA.2.3.7 in Asia is remarkable. Because there are no reliable genomic data from early cases in Malaysia and Vietnam, our phylogenic analysis and the metadata from GISAID suggests that travel between countries in Asia contributed to the rapid spread of this unique Omicron lineage.

In summary, our genomic dataset is uniquely valuable for understanding how a major COVID-19 outbreak occurs in a naive and vaccinated population in Taiwan, a country with a very limited number of entry events. We theorize that the dominant circulation of BA.2.3.7 in Taiwan is likely the result of genetic drift or a founder effect, although it is also possible that increased transmissibility or vaccine evasion played some part. As countries in Asia move from zero tolerance to more open COVID-19 policies, continued surveillance of SARS-CoV-2 using next-generation sequencing is important. Early detection of viral evolution events in endemic areas will help minimize future disruptions caused by a new variant.